Overexpressed cyclin D3 contributes to retaining the growth inhibitor p27 in the cytoplasm of thyroid tumor cells

Abstract

The majority of thyroid carcinomas maintain the expression of the cell growth suppressor p27, an inhibitor of cyclin-dependent kinase-2 (Cdk2). However, we find that 80% of p27-expressing tumors show an uncommon cytoplasmic localization of p27 protein, associated with high Cdk2 activity. To reproduce such a situation, a mutant p27 devoid of its COOH-terminal nuclear-localization signal was generated (p27-NLS). p27-NLS accumulates in the cytoplasm and fails to induce growth arrest in 2 different cell lines, indicating that cytoplasm-residing p27 is inactive as a growth inhibitor, presumably because it does not interact with nuclear Cdk2. Overexpression of cyclin D3 may account in part for p27 cytoplasmic localization. In thyroid tumors and cell lines, cyclin D3 expression was associated with cytoplasmic localization of p27. Moreover, expression of cyclin D3 in thyroid carcinoma cells induced cytoplasmic retention of cotransfected p27 and rescued p27-imposed growth arrest. Endogenous p27 also localized prevalently to the cytoplasm in normal thyrocytes engineered to stably overexpress cyclin D3 (PC-D3 cells). In these cells, cyclin D3 induced the formation of cytoplasmic p27-cyclin D3-Cdk complexes, which titrated p27 away from intranuclear complexes that contain cyclins A-E and Cdk2. Our results demonstrate a novel mechanism that may contribute to overcoming the p27 inhibitory threshold in transformed thyroid cells.

Figure 1

Expression and localization of p27 and cyclin D3 (Cyc D3) in thyroid carcinoma–derived cell lines. (a) Western blot analysis of p27 and cyclin D3 expression in thyroid tumor–derived cell lines. α-tubulin (Tub) was used to assure uniform loading (third row). Bottom row: Cdk2 activity in protein extracts from the same cell lines using histone H1 as substrate. (b) Immunofluorescence analysis of p27 expression, ×100. (c) Western blot analysis of p27 and cyclin D3 on differentially fractionated proteins (C, cytoplasmic fractions; N, nuclear fractions). As control, antibodies against α-tubulin or cyclin H were used. (d) Cdk2 immunoprecipitates: p27 bound to Cdk2, Cdk2 level, and Cdk2 activity in total (left column) or fractionated extracts (remaining columns). (e) Colocalization of p27 and cyclin D3 in thyroid tumor cell lines.

Figure 2

Expression and localization of p27 and cyclin D3 in thyroid tumors. (a) Western blot analysis of p27 or cyclin D3 expression in thyroid tumors. Lane NT, normal thyroid; lanes 1–5, papillary carcinomas; lanes 6 and 7, follicular carcinomas; lanes 8–10, anaplastic carcinomas. α-tubulin was used to assure uniform loading (third row). Bottom, Cdk2 activity in protein extracts from the same samples. (b) p27 immunoreactivity in normal and neoplastic thyroid tissue. Top left: normal thyroid, nuclear p27. Top right: papillary carcinoma, cytoplasmic p27. Bottom left: papillary carcinoma, nuclear and cytoplasmic p27. Bottom right: same tumor with antibody preincubated with antigen. ×40. (c) Western blot analysis of p27 and cyclin D3 on differentially extracted proteins. As control, antibodies to α-tubulin or cyclin H were used. (d) Cdk2 immunoprecipitates: p27 bound to Cdk2, Cdk2 level, and Cdk2 activity in total (left column) or fractionated extracts (remaining columns). (e) Correlation between cyclin D3 expression and p27 localization. Data are from Table 1.

Figure 3

Nuclear localization of p27 is required for p27 activity. (a) Localization of pEGF-p27 and pEGF–p27-NLS in FB-1 cells. Green fluorescence of EGFP was used to identify the localization of fused p27 (top) or was merged with HOECHST staining of nuclei (bottom). (b) BrdU incorporation of FB-1 cells transfected with pEGF-p27 or pEGF–p27-NLS. Left: red arrow shows a pEGF-p27–transfected cell that is not incorporating BrdU. Right: yellow arrow shows a cell transfected with pEGF–p27-NLS incorporating BrdU. (c) BrdU incorporation of Bosc23 cells transfected with p27 or p27-NLS. Green fluorescence of the EGFP protein was used as a flag to identify the localization of fused p27 (top). Middle: BrdU staining. Bottom: HOECHST staining. Left column: red arrows show pEGF-p27–transfected cells that have not incorporated BrdU. Right column: yellow arrow shows a cell transfected with pEGF–p27-NLS and incorporated BrdU. A Neo-Achromat Zeiss objective lens (×100) was used. (d) Statistical analysis of inhibition of BrdU uptake in FB-1 and Bosc23 cells transfected with p27 or p27-NLS. Data represent the mean ± SD of 3 different experiments performed in duplicate; at least 200 cells were counted. (e) Plasmid expression (left) and localization (right) in Bosc23 cells. (f) Cdk2 immunoprecipitates in Bosc23 cells: p27 bound to Cdk2, Cdk2 level, and Cdk2 activity in total or fractionated extracts. E, EGFP; E1, wild-type pEGF-p27; E4, pEGF–p27-NLS.

Figure 4

Cyclin D3 interferes with p27 localization and activity in FB-1 cells. (a) FB-1 cells were transiently transfected as indicated. Localization of p27 and p21 was revealed by green fluorescence. Cyclin D3 was revealed by Texas red–conjugated secondary antibodies. (b) Statistical analysis of p27 and p21 localization in FB-1 cells. Data represent the mean of 3 different experiments performed in duplicate; at least 200 cells were counted. (c) Statistical analysis of the inhibition of BrdU uptake in FB-1 cells by p27 and p21 in the presence or absence of cyclin D3. Data represent the mean of 3 different experiments performed in duplicate; at least 200 cells were counted.

Figure 5

Cyclin D3 interferes with p27 localization and activity in Bosc23 cells. (a) Mutant p27 constructs. (b) Expression of mutant constructs in Bosc23 cells. (c) Immunofluorescence analysis of p27 localization in Bosc23 cells transfected with the indicated constructs. (d) Immunofluorescence analysis of p27 expression and localization in Bosc23 cells transfected with the indicated construct. Green fluorescence of pEGF-fused proteins was overlaid on the phase-contrast photos to reveal localization. (e) Relative efficacy of cyclin D3 overexpression to sequestrate different p27 mutant constructs in the cytoplasm. (f) Relative efficacy of cyclin D3 overexpression to reverse inhibition of BrdU incorporation induced by the different p27 mutant constructs. E1, p27 (wild-type); E4, p27-NLS; E2, p27-TA187; E3, p27-97-197; E5, p27-1-186.

Figure 6

Cyclin D3 induces p27 retention in the cytoplasmic compartment in PC-D3 cells. (a) Western blot analysis of p27 and cyclin D3 in PC Cl 3 and PC-D3 cells. Total cell lysates (T), cytoplasmic fractions (C), and nuclear fractions (N) were analyzed. (b–d) Analysis of cytoplasmic and nuclear complexes. Immunoprecipitates of (b) cyclin D3, (c) cyclin E, and (d) cyclin A. (e) Immunofluorescence analysis of p27 in PC Cl 3 and PC-D3 cells. Cells were stained with anti-p27 antibodies and revealed with Texas red–conjugated antibodies (top). Cells were stained with anti–cyclin D3 antibodies (first and second columns) or anti-HA (third and fourth columns) and revealed with FITC-conjugated antibodies (II Ab; second row). Overlapping of red and green staining is shown in the third row. HOECHST staining of cell nuclei appears in the fourth row. Each experiment was repeated 3 times; at least 200 cells were counted. Statistical results for p27 nuclear localization and BrdU incorporation are reported at the bottom. ×100.